Cloth Simulation Getting Eaten By Mesh

So I’m making a model for animation, and I wanted it to have some clothes that are all flowy. I modeled the body and one piece of clothing so far, and it looks fine barring one glaring issue:


This happens.

I’ve tried increasing quality steps and the thickness of the collider.
Increasing quality has no effect outside slowing the program down and increasing the thickness of the body just makes it worse.
I’ve tried smoothing the collider, smoothing the body, doing both and adding more verts and smoothing them again, and no matter what, the cloth wants to get eaten by the body.
I’ve also tried to use another 3rd party cloth solution and it behaves the same way so I figure I must be doing something terribly wrong. If someone could point me in the right direction on this I’d be very grateful.

I guess the scale is far to big. Your model look like 100m tall. That make calculations unprecise.
Try at a smaller realistic scale. Experiment with the values just with a sphere and layed an cloth on it.

I might be wrong being fairly new to blender and 3d modeling in general, but it doesn’t seem big at all based on the positional coords. In either case, I tried scaling to 1/100th of its size and it seems to exacerbate the issue, even if I apply the transform to the meshes

Oh I forgot to talk about the sphere test, it seems to work perfectly fine with it.

After resizing you must apply the size, object -> apply -> scale or use crtl + a. This should solve the exacerbate issue. It not try to apply all transforms

I did it both ways in my testing

https://cdn.discordapp.com/attachments/507298799169830941/645340508301819925/blend2.mp4 After a good deal of testing this is where I am. All objects are using default settings. It only breaks on my mesh and I don’t know why.

If you’d like to mess with the original file, here it is:
https://mega.nz/#!VDQmkAAS!3587dMs7VxeCnByhm2_BBpVTU3OdhmpdLVqAJD_Dm7c

Alright so after around 8 hours of trying to figure this out, I’ve done it but I don’t know why.

The normals apparently got screwed up at some point. I had to recalculate the inside, and then the outside, and flip the normals after that.

And now it works perfectly.
I don’t know why this works, but I’m thankful that it does.